User:Branchiobdellid/Ecological forecasting

Methods
Ecologists shifted towards Bayesian methods starting 1990, when improvements in computational power allowed the use of more demanding computational statistics such as Hierarchical Bayes. This kind of analysis employs a Bayesian Network that provides a probabilistic graphical model of a set of parameters, and can accommodate unobserved variables. A Bayesian structure is a probabilistic approach that is flexible for high-dimensional data, and allows ecologists to separate sources of uncertainty in their models.

Forecasts can leverage Bayes' Theorem and be iteratively updated with new observations using a process called Data Assimilation. Data Assimilation combines observations on different temporal and geographic scales with forecasts, all of which combine to provide more information than any one data source alone. Some ecologists have found this framework to be useful for ecological models as they often rely on a wide range of data sources.